Power converter; power electronics; semiconductor switches; electrical ... An ideal static converter allows control of the power flow between the two sources with ...
Definition of Power Converters F. Bordry and D. Aguglia CERN, Geneva, Switzerland Abstract The paper is intended to introduce power conversion principles and to define common terms in the domain. The concepts of sources and switches are defined and classified. From the basic laws of source interconnections, a generic method of power converter synthesis is presented. Some examples illustrate this systematic method. Finally, the commutation cell and soft commutation are introduced and discussed. Keywords Power converter; power electronics; semiconductor switches; electrical sources; design rules; topologies. 1
Introduction
The task of a power converter is to process and control the flow of electrical energy by supplying voltages and currents in a form that is optimally suited for user loads. Energy conversions were initially achieved using electromechanical converters (which were mainly rotating machines). Today, with the development and the massive production of power semiconductors, static power converters are used in numerous application domains and especially in particle accelerators. Their weight and volume are smaller and their static and dynamic performance are better. A static converter is composed of a set of electrical components building a meshed network that acts as a linking, adapting, or transforming stage between two sources, generally between a generator and a load (Fig. 1).
Fig. 1: Definition of a power converter
An ideal static converter allows control of the power flow between the two sources with 100% efficiency. A large part of power converter design is the optimization of its efficiency. But as a first approach and to define basic topologies, it is interesting to take the hypothesis that no losses occur
through a power converter’s conversion process. With this hypothesis, the basic elements are of two types:  nonlinear elements that are, most of the time, electronic switches: semiconductors used in commutation mode [1];  linear reactive elements: capacitors, inductances (and mutual inductances or transformers). These reactive components are used for intermediate energy storage but also perform voltage and current filtering. They generally represent an important part of the size, weight and cost of the equipment [2, 3]. The objective of this introductory paper is to recall and give a precise definition of basic concepts essential for the design or understanding of power converter topologies [4]. First, a very simple example is presented to illustrate the basic principle of modern power electronics converters (switching power converter). Then the sources and the switches are defined, followed by the fundamental connection rules between these basic elements. From there, converter topologies are deduced. Some examples of topology synthesis are then given. Finally, the concept of hard and soft commutation is introduced. 2
The very basic principles of modern power electronics conversion
As a first basic example to illustrate the evolution toward modern power electronics, let us consider a DC to DC converter that aims to deliver 100 V to a resistive load of 10 Ω. The input voltage source delivers a constant 325 V. Until the 1960s, and still in use in some special applications, a typical utilization of transistors consisted of operating them in their linear, or active, region. The basic ‘old’ topology for this case is illustrated in Fig. 2. This consists of operating the transistor such that a 225 V voltage drop across it is ensured. Simplifying, one can say that in this case the transistor is used as a controllable resistance.
Fig. 2: DC to DC converter using the transistor’s linear, or active, region
Analysing this circuit at this operating point, one can derive the input and the delivered output powers Pin and Pout, the losses in the transistor PT, and the converter efficiency η, as presented in Eq. (1), 𝑃𝑃in = 325 V × 10 A = 3.25 kW ,
(1)
𝑃𝑃T = 𝑃𝑃in − 𝑃𝑃out = 2.25 kW ,
(3)
𝑃𝑃out = 100 V × 10 A = 1 kW , 𝜂𝜂 =
𝑃𝑃out 𝑃𝑃in
= 0.3 ≡ 30% .
(2)
(4)
Notice that for this particular operating point the efficiency is extremely low. In the general case, the efficiency can be expressed as in Eq. (5), where the efficiency drastically decreases as Vout decreases, 𝜂𝜂 =
𝑃𝑃out 𝑃𝑃in
=
2 𝑉𝑉out � 𝑅𝑅 (𝑉𝑉in ∙𝑉𝑉out )� 𝑅𝑅
=
𝑉𝑉out 𝑉𝑉in
.
(5)
This conversion system produces high losses and needs an overdimensioned transistor able to dissipate the losses. Furthermore, the overall dimensions of this power converter shall be large enough
for being able of evacuating the high losses for a given maximum temperature. This conversion method is still used in some special applications where high precision or high dynamics is required [5]. The basic principle of modern power electronics lies in the utilization of switches in their ON and OFF states only, virtually producing no, or very low, losses (losses in switches are described in Section 4.2, Dynamic characteristics). The basic principle is illustrated in Fig. 3. Between the voltage source Vin and the load R, the converter is now composed of a ‘special’ twoposition switch and a lowpass filter. When the switch is in position 1, the input voltage Vin appears at the lowpass filter input Vs. When the switch is in position 2, zero volts are applied to Vs. Given a pattern in time of the switch positions, one derives the pattern of the voltage Vs (chopped or switched voltage). In most applications, the voltage Vs, with a fluctuation of 100% (from 0 V to Vin), cannot be directly applied to the load, therefore a filtering action must be undertaken in order to apply the voltage Vs average values to the load, which is derived in Eq. (6), 𝑉𝑉out =
1 𝑇𝑇s ∫ 𝑣𝑣𝑖𝑖𝑖𝑖 (𝑡𝑡)d𝑡𝑡 𝑇𝑇s 0
= 𝐷𝐷𝑉𝑉in ,
(6)
where Ts is the switching period (which defines the switching frequency fs), and D is the duty cycle, or duty ratio, which defines the relative time when the switch is in position 1 with respect to the switching period Ts (illustrative definitions are given in Fig. 3). In this case, the regulation of the output voltage is performed by acting on the duty cycle D.
Fig. 3: Idealized switchmode DCtoDC converter principle
Thanks to the low losses, modern switchmode power converters are more efficient and compact. However several new aspects have to be considered during the design phase. A filtering process is required, leading to the introduction of the socalled output voltage and current ripples on the load and voltage, or current, bandwidths. A nonideal filter is always letting some harmonics pass through the load. Furthermore, the dynamic characteristics of the power converter (rate of change of current and/or voltage) is limited by the filter. This tradeoff between ripple and bandwidth is a key aspect in the design and specification process of modern switchmode power converters. 3
Sources
As mentioned in the introduction, a power converter processes the flow of energy between two sources. To synthesise a power converter topology, the first step is to characterize these sources. We will see later that the converter structure can be directly deduced as soon as the sources are defined: voltage or current sources and their reversibilities.
In the energy conversion process, a source is usually a generator (often called an input source) or a load (often called an output source). However, in the case of a change of direction of the energy flow, i.e. a change in the sign of the power, the sources (generators and loads) can exchange their functions (i.e. restoration of energy from a magnet back to the grid). 3.1 Nature of sources 3.1.1
Definitions
Two types of sources could be defined: voltage and current sources. As mentioned, any of these sources could be a generator or a receiver (load). A source is called a voltage source if it is able to impose a voltage independently of the current flowing through it. This implies that the series impedance of the source is zero (or negligible in comparison with the load impedance) A source is called a current source if it is able to impose a current independently of the voltage at its terminals. This implies that the series impedance of the source is infinite (or very large in comparison with the load impedance). These definitions correspond to permanent properties. The principle of operation of a converter is based on the switchmode action of its switches. Commutation of the switches generates very fast current and/or voltage transients so that the transient behaviour of the sources is of fundamental importance for the converter design. The transient behaviour of a source is characterized by its ability or inability to withstand steps in the voltage across its terminal or in the current flowing through it, these steps being generated by the external circuit. Then new definitions could be stated as below.  A source is a voltage source if the voltage across its terminals cannot undergo a discontinuity due to the external circuit variation. The most representative example is a capacitor, since an instantaneous change of voltage would correspond to an instantaneous change of its charge that would require an infinite current (Fig. 4(b)).  A source is a current source if the current flowing through it cannot undergo a discontinuity due to the external circuit variation. The most representative example is an inductor, since an instantaneous change in current would correspond to an instantaneous change in its flux that would require an infinite voltage (Fig. 4(a)).
Fig. 4: Inductance and capacitor vs. current and voltage source
It should be noted that a square wave voltage generator is indeed a voltage source as defined above since the voltage steps are not caused by the external circuit. A square wave current generator is indeed a current source as defined above since the current steps are not caused by the external circuit. With these definitions, it is interesting to define the notion of instantaneous impedance of a source as the limit of the source impedance when the Laplace operator tends towards infinity. Theoretically this instantaneous impedance can be zero, finite, or infinite.
A source is referred to as a voltage source when its instantaneous impedance is zero, while a source is called a current source if its instantaneous impedance is infinite. For example: Capacitor:
Z(s) = 1/(C.s) , limit
s→∞
Inductance: Z(s) = L.s ,
limit
s→0
Z(s) = 0, this leads to a voltage source; Z(s) = ∞, this leads to a current source.
3.1.2 Source reversibility The determination of the source reversibilities is fundamental. We will see that the reversibility analysis allows the deduction of the static characteristics of the switches. The voltage (or the current) that characterizes a source is termed DC if it is unidirectional. As a first approximation, it can be taken as being constant. The voltage (or the current) is termed AC if it is periodic and has an average value equal to zero. As a first approximation, it can be taken as sinusoidal. A source is voltagereversible if the voltage across its terminals can change sign. In the same way, a source is currentreversible if the current flowing through it can reverse. In summary, the input/output of a converter can be characterized as voltage or current sources (generator or loads), either DC or AC, currentreversible and/or voltagereversible. In total, there are only eight possibilities, shown in Fig. 5.
Fig. 5: Voltage and current sources with their reversibilities
3.1.3
Source nature modification
Connection of a series inductance with an appropriate value to a voltage source (i.e. a dipole with zero instantaneous impedance) turns the voltage source into a current source. In the same way, connecting a parallel capacitor of appropriate value to a current source (a dipole with infinite instantaneous impedance) turns the current source into a voltage source (Fig. 6).
Fig. 6: Source nature confirmation or modification
These inductive or capacitive elements connected in parallel or in series with the source are elements that can temporarily store energy. Consequently, if an inductance connected to a voltage source turns it into a current source it is important to determine the current reversibility of this current source. In practice the identification of a real generator or of a real load as a voltage or current source is not obvious. That is the reason why the nature of the source is often reinforced by the addition of a parallel capacitor in the case of voltage sources and by the addition of a series inductor in the case of current sources. Obviously, the current source obtained by connecting an inductance in series with a voltage source keeps the same current reversibility as this voltage source. The inductance acts as a buffer absorbing the voltage differences. Consequently the current source obtained by connecting a series inductance to a voltage source is reversible in voltage. When the voltage source itself is reversible in voltage there is no particular problem. But, if the voltage source is not reversible in voltage, the current source obtained by connecting a series inductance to the voltage source is only instantaneously reversible with respect to voltage. The former result can easily be transposed to the voltage source obtained by the parallel connection of a capacitor to a current source. The voltage source obtained keeps the same voltage reversibility as the current source and is reversible in current. However, this reversibility is only instantaneous if the current source is not reversible in current. 3.1.4
Example
A set of ideal batteries behaves as a load during charging and as a generator during discharging; such a source is called a DC voltage source, being current reversible but not voltage reversible. Nevertheless, because of the inductance of the connecting cables, this battery can sometimes be taken as a current source that is instantaneously voltage reversible and permanently current reversible. If a capacitor bank is added at the terminals of the cables, it again becomes a voltage source (Fig. 7).
Fig. 7: Modification to voltage or current source
4
Switch characteristics
Static converters are electrical networks mainly composed of semiconductor devices operating in switch mode (as switches). Through proper sequential operation of these components, they allow an energy transfer between two sources with different electrical properties. The losses in the switches should be minimized in order to maximize the efficiency of the converter. Switches must have a voltage drop (or an ON resistance) as low as possible in the ON state, and a negligible leakage current (or an OFF resistance) in the OFF state. These two states are defined as static states.
The change from one state to the other state (switch commutation) implies transient behaviour of the switch. These behaviours are complex because they depend on the control of the switch (through a gate control) and on the conditions imposed by the external circuit. 4.1 Static characteristics In the static domain a switch has the same behaviour as a nonlinear resistance: very low in the ON state and very high in the OFF state. Taken as a dipole with the load sign convention (Fig. 8) the static characteristic Ik(Vk), which represents the operating points of a switch, is made up of two branches totally located in quadrants 1 and 3 such that (Vk × Ik) > 0. One of these branches is very close to the Ik axis (ON state) and the other is very close to the Vk axis (OFF state). Each of these branches can be located in one or two quadrants. In the case of an ideal switch, the static characteristics are the halfaxis to which they are close.
Fig. 8: Static characteristics of a switch
In this representation, except for the obvious cases of a short circuit and of an open circuit that correspond respectively to a switch always ON and to a switch always OFF, any switch that really behaves as a switch (commutation: ON OFF) has a static characteristic consisting of at least two orthogonal halfaxes (or segments). The static characteristic, an intrinsic feature of a switch, reduces to a certain number of segments in the Ik(Vk) plane. –
Twosegment characteristics: the switch is unidirectional in current and in voltage. Two twosegment characteristics can be distinguished: in the first case, current Ik and voltage Vk have the same signs; in the second case, current Ik and voltage Vk have opposite signs. The switches having such characteristics are respectively called T and D switches (Fig. 9).
Fig. 9: Static characteristics of a twosegment switch
–
Threesegment characteristics: the switch is bidirectional either in current or in voltage while the other is unidirectional. Therefore, there are two types of threesegment static characteristics (Fig. 10). It should be noted that these two types of switches could be synthesized with the association in parallel or in series of twosegment switches (T and D).
Fig. 10: Static characteristics of a threesegment switch
–
Foursegment characteristics: the switch is bidirectional in voltage and in current. There is only one such type of static characteristic (Fig. 11). A foursegment characteristic could be obtained by series or parallel connection of switches with threesegment characteristics.
Fig. 11: Static characteristics of a foursegment switch
4.2 Dynamic characteristics The dynamic characteristic is the trajectory described by the point of operation of the switch during its commutation, to go from one halfaxis to the perpendicular halfaxis. A switch being either ON or OFF, there are two commutation dynamic characteristics corresponding to the turnON and the turnOFF, which will be grouped under the global term dynamic characteristics. Unlike the static characteristic, the dynamic characteristic is not an intrinsic property of the switch but also depends on the constraints imposed by the external circuit. Neglecting secondorder phenomena, and taking into account the dissipative nature of the switch, the dynamic characteristic can only be located in those quadrants where Vk × Ik > 0 (generator quadrants). For the two commutations (turnON and turnOFF), two modes are possible: controlled commutation and spontaneous commutation. 4.2.1
Controlled commutation
The switch has, in addition to its two main terminals, a control terminal on which it is possible to act in order to provoke a quasiinstantaneous change of state (in the case of a T switch). The internal resistance
of this switch can change from a very low value to a very high value at turnOFF (and inversely at turnON). These changes are independent of the evolution of the electrical quantities imposed on the switch by the external circuit. It should be noted that, in a controlled commutation, the switch imposes its state on the external circuit. Under such circumstances, the element can undergo severe stresses that depend on its dynamic characteristic. If the switching time is long and the operating frequency is high, the commutation losses can be important. 4.2.2
Spontaneous commutation
The spontaneous commutations correspond to turnOFF when the current flowing through the switch arrives at zero and to turnON when the voltage applied across its terminals reaches zero. Spontaneous commutation is the commutation of a simple PN junction (D switch). It is only dependent on the evolution of the electrical variables in the external circuit. Spontaneous commutations could be achieved with any controlled semiconductor if the gate control is synchronized with the electrical quantities of the external circuit. Spontaneous commutation is achieved with minimal losses since the operating point moves along the axes. It is important to point out that controlled commutation can only happen in the first or third quadrants while spontaneous commutation can only happen with a change of quadrant (Fig. 12).
Fig. 12: Spontaneous and controlled commutations
4.3 Classification of switches Finally, switches used in power converter can be classified by their static characteristics (two, three or four segments) and by the type of commutation (controlled or spontaneous) at turnON and at turnOFF. 4.3.1
Twosegment switches
Except the open circuit and the short circuit, two switches with twosegment characteristics can be distinguished (Fig. 13). –
The first of these switches has the static characteristics of switch D, and its turnON and turnOFF commutations are spontaneous. This switch is typically a diode.
–
The second of these switches has the static characteristics of switch T and its turnON and turnOFF commutations are controlled. Examples are the power semiconductors: metaloxide semiconductior fieldeffect transistors (MOSFET), insulatedgate bipolar transistor (IGBT), gate tunroff thyristor (GTO), integratedgate commutated thyristor (IGCT), etc.
Fig.13: Dynamic characteristics of twosegment switches
This switch will be symbolized by separating the turnON and turnOFF control gate as shown in Fig. 13. A switch with twosegment characteristics similar to the T type (both segments in the same quadrant), must have controlled turnON and turnOFF commutations. If it would have only one controlled commutation, it would be necessary to put in series or in parallel a switch D (a diode) to get the spontaneous commutation. In this case, it is no longer a twosegment switch, but a threesegment one. Therefore, only two twosegment switches can be used directly. 4.3.2
Threesegment switches
These switches can be divided into two groups depending whether they are: –
unidirectional in current and bidirectional in voltage (Fig. 14);
–
bidirectional in current and unidirectional in voltage (Fig. 15).
Fig, 14: Dynamic characteristics of threesegment switches: bidirectional in voltage
Fig. 15: Dynamic characteristics of threesegment switches: bidirectional in current
Except for the thyristor, all these switches are synthesized switches, realized by connecting a diode in parallel or in series with a twosegment switch. A special driver is needed to obtain spontaneous commutation. The ‘dualthyristor’ (unidirectional in voltage, bidirectional in current, controlled turnOFF and spontaneous turnON) is a good example of a useful threesegment switch [6, 7]. In each of these two groups, all switches have the same static characteristics but differ with their commutation mechanisms. It is important to remark that if a threesegment switch has both commutations controlled (turnON and turnOFF) or both spontaneous, it would never use the three segments of its static characteristic. Therefore, a threesegment switch must necessarily have one controlled commutation and one spontaneous commutation. The cycle of operation, which represents the locus described by the point of operation of these switches, is then fully determined. They can only be used in converter topologies that impose a single cycle on the switches during operation. 4.3.3
Foursegment switches
All foursegment switches have the same static characteristic. They differ only by their commutation modes that can, a priori, differ in quadrants 1 and 3. So, six foursegment switches can be distinguished. These switches are used mainly in direct frequency changers and in matrix converters; in practice they are made up of two threesegment switches connected in series or parallel. 5
Interconnection of sources: Commutation rules
To control the power flow between two sources, the operation principle of a static power converter is based on the control of switches (turnON and turnOFF) with determined cycles creating periodic modifications of the interconnection between these two sources. The source interconnection laws can be expressed in a very simple way: –
a voltage source should never be shortcircuited but it can be opencircuited;
–
a current source should never be opencircuited but it can be shortcircuited.
From these two general laws, it can be deduced that switches cannot establish a direct connection between two voltage sources or between two current sources.
Fig. 16: Basic interdictions of source interconnection
In the case of two voltage sources, the switch turnON can only happen when the two sources have the same values, that is to say at the zero crossing of the voltage across the switch. The turnON must then be spontaneous (since it depends on the external circuit) and turnOFF can be controlled at any time. In the case of two current sources, the switch turnOFF can only happen when the two current sources reach the same value, that is to say when the current in the switch reaches zero. In this case, the turnOFF is spontaneous and turnON can be controlled at any time. As it is done in daytoday practise and following the previous arguments, it is obvious that it is possible to put capacitors in parallel and inductances in series but these elements should be connected with zero voltage or zero current, respectively. From the laws state above, it is easy to deduce that the commutation of switches must not interconnect two sources of the same nature. 6
Structure of power converters
A power converter can be designed with different topologies and with one or several intermediate conversion stages. When this conversion is achieved without any intermediate stage temporarily storing some energy, the conversion is called direct conversion and it is achieved by a direct converter. On the other hand, when this conversion makes use of one or more stages able to store energy temporarily, the conversion is termed indirect and it is achieved by an indirect converter. The interdiction to connect two sources of the same nature leads to the consideration of two classes of basic conversion topologies: –
direct link topology: when the two sources have different natures;
–
indirect link topology: when the two sources have the same nature.
6.1 Direct link topology converters A direct converter is an electrical network composed only of switches and is unable to store energy. In such a converter, the energy is directly transferred from the input to the output (with the hypothesis that the losses can be neglected); the input power is at any time equal to the output power. Taking into account the interconnection rules recalled above, the different possible connections between a voltage source and a current source are shown in Fig. 17. To get all these connections, the simplest structure is the fourswitches bridge (Fig. 18): –
with K1 and K3 closed, the connection shown in Fig. 17(a) is obtained;
–
with K2 and K4 closed, the connection shown in Fig. 17(b) is obtained;
–
with K1 and K2 closed (or with K3 and K4 closed), the connection shown in Fig. 17(c) is obtained.
(a)
(b)
Connection (energy flow between sources)
(c)
Disconnection (Current source shortcircuited, voltage source opencircuited)
Fig. 17: Interconnection possibilities between a voltage and a current source. (a) – direct connection; (b) – inverse connection; (c)  disconnection with current source in shortcircuit.
Fig. 18: Basic configuration of voltage–current direct converter
When some of these connections are not necessary, it is possible to simplify the bridge structure into structures using fewer switches (i.e. a buck converter). Energy conversion between an input current source and an output voltage source is the same problem. The basic configuration is the same but it is more usual to represent the input source on the left and the output source on the right (Fig. 19).
Fig. 19: Basic configuration of current–voltage direct converter
6.2 Indirect converters If both sources have the same nature, it is not possible to interconnect them directly with switches. It is necessary to add components to generate an intermediate buffer stage of a different type without active energy consumption (capacitor or inductor). This buffer stage is a voltage source (capacitor) if the energy transfer is between two current sources, and it is a current source (inductance) if the energy transfer is between two voltage sources.
6.2.1
Modification of the nature of the input or output source
In the case of voltage–voltage conversion, one solution could be to add an inductance in series with the input voltage source or with the output voltage source. With this change of the source nature, it is possible to use a direct converter: a current–voltage or voltage–current converter according to where the inductance was added (Fig. 20).
Fig. 20: Indirect converters: modification of the nature of the input or output source. Above one of the voltage sources need to be modified into a current source. Below one of the current sources need to be modified into a voltage source.
The case of current–current conversion is similar to the previous case. Therefore, a capacitor should be added in parallel or in series with the input current source or with the output current source. 6.2.2
Use of two direct converters
If it is not possible or too costly to modify the nature of one source, two direct converters can be connected with an intermediate buffer stage: an inductance for a voltage–voltage conversion and a capacitor for a current–current conversion (Fig. 21).
Fig. 21: Indirect converters: intermediate buffer stage, inductive above and capacitive below, between two direct converters.
6.2.3
Voltage–voltage indirect converters
In indirect converters, the two voltage sources are never connected directly (see the basic laws described above). Two sequences are therefore necessary. During the first sequence, the energy is transferred from the input voltage source to the inductance (voltage to current conversion). During the second sequence, the inductance gives back the energy to the output voltage source (current to voltage conversion; two directions are possible) (Fig. 22). An extra switch is necessary to obtain these sequences.
Fig. 22: Voltage–voltage indirect converters. Above the three possible connections between first and second sequences. Below the corresponding topology.
6.2.4
Current–current indirect converters
In the indirect converters, the two current sources are never connected directly. Then, two sequences are necessary. During the first sequence, the energy is transferred from the input current source to the capacitor (current to voltage conversion). During the second sequence, the capacitor gives back the energy to the output current source (voltage to current conversion; two directions are possible) (Fig. 23).
Fig. 23: Current–current indirect converters. Above the three possible connections between first and second sequences. Below the corresponding topology.
6.3 Conclusions Figure 24 represents the three basic configurations of all singlephase converters.
Fig. 24: Three basic structures of singlephase power converters. Voltage to current (and viceversa) direct conversion (above), voltage to voltage indirect conversion (in the middle), and current to current indirect conversion (below).
From these basic configurations, it is possible, according to the nature of the sources, to associate them or to add other components. For example, in the case where one of the sources is AC, it is possible to insert a transformer for adaptation or galvanic insulation (Fig. 25).
Fig. 25: Insertion of a transformer in a direct topology
It is also possible to associate several basic topologies. One application that is more frequently used is to have an intermediate stage operating at a high frequency to reduce the size of the transformers and magnetic elements and to get higher performance at the output (bandwidth, ripple, etc.) (Fig. 26). Details of magnetic component design are given in Ref. [3].
Fig. 26: Association of elementary structures
Figure 27 illustrates the case of resonant converters [8, 9]. It should be noted that the interconnection of the various intermediate sources must respect the interconnection laws. It is especially important for the choice of the output filter.

+
I L Voltage
C
Current
Voltage
Current

+
V

C
L Voltage
Current
Voltage
Current
Fig. 27: Series (above) and parallel (below) resonant converters
7
Power converter classification
Figure 28 is a table summarizing the power conversion topologies. The crossed cells correspond to reversibility incompatibilities between the input and output sources. Two symmetric cells, with respect to the diagonal line D, represent two reversible topologies. Two topologies corresponding to two cells symmetric with respect to the point O are dual.
Fig. 28: General power conversion table
Figure 29 gives the different types of power converters and their usual names.
Fig. 29: Power converter classification
8
Synthesis of power converters
8.1 Synthesis method A general and systematic method to obtain the power converter topologies is given below. 1. Determine the natures of the input and output sources. The basic structure can then be chosen (Fig. 24). 2. From the specification, deduce the voltage and current reversibilities of the input and output sources (one configuration from the general table (Fig. 28)). 3. From the basic structure, identify the different phases of operation according to the reversibilities and the energy transfer control. If necessary, simplify the configuration (shortcircuit or open switches). 4. For the various phases, check the sign of the current through the ON switches and the sign of the voltage across the OFF switches: the static characteristics Ik(Vk) of each switches are defined. 5. From the specification (desired output current and/or voltage functions), deduce the sequence of the different phases. For every commutation, represent the working point of each switch before and after the commutation. 6. From the static and dynamic characteristics, choose the type of switches (semiconductor type). 8.2 Examples 8.2.1
Nonreversible current chopper
Hypothesis: –
power conversion between a voltage source and a current source;
–
these two sources are unidirectional in voltage and in current. Following the steps of the synthesis method:
1. a direct converter topology can be used; 2. cell (1,8) of the general table according to the source reversibilities; 3. Sequence 1 (active phase) and 2 (free wheel phase) from Fig. 30; 4. the different plots for the four switches for the two sequences are represented in Fig. 30; 5. from the previous plots, it can be easily deduced that: –
K1 is a controlled twosegment switch (T switch) (Fig. 13);
–
K2 is an inverse D switch;
–
K3 is a shortcircuit;
–
K4 is an opencircuit.
The topology of the converter is thus fully determined.
(a)
(c)
(b)
Fig. 30: Study of a nonreversible current chopper. (a) – sequences 1 and 2 (change in K1, K2, and K4); (b) – Currentvoltage characteristics of each switch for the two sequences; (c) corresponding topology with transistor and diode.
8.2.2 Reversible current chopper Hypothesis: –
power conversion between a voltage source and a current source;
–
the input voltage source is bidirectional in current;
–
the output current source is unidirectional in voltage and bidirectional in current. Following the steps of the synthesis method:
1. a direct converter topology can be used; 2. cell (2,6) of the general table according to the source reversibilities; 3. sequence 1 (active phase) and 2 (free wheel phase) of Figs. 30 and 31; 4. the different plots for the four switches are represented in Fig. 30 for the two sequences for the transfer of energy from the input source to the output source. Figure 31 represents the analysis of the brake phase: transfer of energy from the output source to the input source. 5. the analysis of the two phases leads to the determination of the structure presented in Fig. 32.
(a)
(c)
(b)
Fig. 31: Study of a reversible current chopper: brake phase. (a) – sequences 1 and 2 (change in K1, K2, and K4); (b) – Currentvoltage characteristics of each switch for the two sequences; (c) corresponding topology with transistor and diode.
Fig. 32: Structure of reversible current chopper
8.2.3 Voltage inverter Hypothesis: –
power conversion between a voltage source and a current source;
–
the input voltage source is unidirectional in voltage and bidirectional in current; it is a DC source of value E.
–
the output current source is bidirectional in voltage and bidirectional in current; it is an AC source. The specification is to get a +E, −E voltage at the output of the converter. The steps of the synthesis method are followed.
1. A direct converter topology can be used. 2. Cell (2,5) of the general table according to the source reversibilities. 3. The two sequences are shown in Fig. 33. 4. From the analysis of two sequences, it can be deduced that the switches must be bidirectional in current and unidirectional in voltage. They correspond to threesegment switches (D and T in parallel; Fig. 10). 5. To represent the working point of each switch, it is necessary to detail the specification. Two cases are defined according to whether the output voltage is ahead or not of the output current.
Fig. 33: Sequence of a voltage inverter
8.2.3 .1 Case 1: The output voltage is ahead of the output current For the hypothesis that the output voltage is ahead of the output current, Fig. 34 shows the dynamic characteristics based on the analysis of the commutation between the two sequences. It can be easily deduced that the four switches (K1, K2, K3, and K4) should be controlled turnOFF and spontaneous turnON switches (dual thyristor; Fig. 15).
(a)
(b)
(c)
Fig. 34: Voltage inverter (Case 1): (a) – output voltage and current; (b) – Switches states for each time period (1a, 1b, 2a, and 2b); (c)  dynamic characteristics of the switches.
The topology of the converter is fully determined and is shown in Fig. 35. It is a zerovoltageswitching topology (see Section 9).
Fig. 35: Voltage inverter (Case 1): converter topology
8.2.3 .2 Case 2: The output current is ahead of the output voltage Identical study can be made if the output current is ahead of the output voltage. Figures 36 and 37 represent the dynamics characteristics of the switches and the deduced topology. It is a zerocurrentswitching topology using a thyristor with a reverse diode in parallel (Fig. 15).
(a)
(b)
(c)
Fig. 36: Voltage inverter (Case 2): (a) – output voltage and current; (b) – Switches states for each time period (1a, 1b, 2a, and 2b); (c)  dynamic characteristics of the switches.
Fig. 37: Voltage inverter (Case 2): converter topology
9
Commutation cell
As presented in the previous chapters, the operation of a static converter can be split into sequences. A distinctive electrical network characterizes each sequence. The interconnection modifications of the sources by switches give the electrical networks. In the general case, the modifications are made by switches, which connect n branches to one. Figure 38 shows an example of a threeway switch.
Fig. 38: Threeway switch
The network branches connected to these switches must fulfil the connection laws of the sources. Therefore, it can be deduced that: –
each switch is connected to a voltage source (otherwise opening a switch would result in opencircuiting a current source);
–
the node at the centre of the star is connected to a current source since a voltage source can be connected only to a current source through a controlled switch;
–
at a given time one and only one switch must be ON to avoid connecting two voltages sources and opencircuiting the current source.
Following these deductions, each commutation mechanism is a sequence of commutation only involving two switches. Thus, an elementary commutation cell, represented in Fig. 39, could be defined. The reversibilities of the voltage and current sources determine the static characteristics of the two switches. The switches need to have static characteristics with the same number of current and voltage segments. The two switches must be complementary, that is to say if one switch is ON the other one is OFF and, furthermore, if the turnON (turnOFF) commutation of a switch is controlled, the turnOFF (turnON) commutation of the other must be spontaneous.
Fig. 39: Elementary commutation cell
To study a power converter topology, it is always fundamental to isolate all the elementary commutation cells and to check the complementarities of the switches. Figure 40 represents several examples of power converter topologies for which the elementary cells were highlighted. Detailed information on commutation cells and on local and system commutation mechanism can be found in Refs. [9, 10, 13].
(a)
(b)
(c)
(d)
(e)
(f)
Fig. 40: Elementary cells in power converter examples. (a) – Chopper; (b) – indirect converter with DC voltage link; (c) – single phase voltage source inverter; (d) – single phase currentsource inverter; (e) – threephase voltagesource inverter; (f) – three phase currentsource inverter.
10
Hard and soft commutation
In the domain of power conversion, the only available active components were diodes and thyristors. The topologies had to respect the commutation of these components: i.e. spontaneous turnOFF of thyristors. The commutation mechanism was called ‘natural commutation’. If the spontaneous turnOFF was not directly fulfilled (especially in the case of DC sources), it was necessary to add auxiliary circuits with reactive components (inductors and capacitors) and auxiliary semiconductors. The goal of these circuits was to create the conditions for thyristor turnOFF. This commutation mechanism was called ‘forced commutation’. Power specialists were dreaming of having a power semiconductor with a controlled turnOFF to avoid adding complex and costly auxiliary circuits and being limited to the grid frequency. With the development of the power transistor and the GTO, a new spectrum of topologies was opened. Then, specialists pushed to get faster components with simple and lighter drivers. From 1985, these components are available on the market. Power converters with higher power and frequency were then designed and built. However, the consequence is the management of (very) high dV/dt and dI/dt with stress on semiconductors but also on all of the other components. EMC became a constant concern of power electronics designers. Hard commutation was born! A first approach was to add components (snubber) to slow down the commutation (series inductor for a controlled turnON and a parallel capacitor for a controlled turnOFF) and to avoid as much as possible the commutation losses close to the axes: aidedcommutation. To avoid discharging the stored energy of the snubber in the
semiconductor at the next commutation (inductance energy at turnOFF and capacitor energy at turnON), it is necessary to add an auxiliary circuit to discharge the snubber energy and then to generate losses. The solution to avoid all these problems and be able to use these new fast semiconductors with lossless snubbers is to have only one controlled command per switch. One commutation is spontaneous and the other one is controlled. Two types of commutations are possible: zerovoltage switching (ZVS; Fig. 41) and zerocurrent switching (ZCS; Fig. 42).
Fig. 41: Zerovoltage switching principle
Fig. 42: Zerocurrent switching principle
The spontaneous commutation is lossless and it is easy to limit the losses for the controlled commutation with a series inductor for ZCS and a parallel capacitor for ZVS. These reactive
components are naturally discharged for the next commutation, which is a spontaneous commutation. This commutation mechanism is called soft commutation. For a power converter topology to obtain the conditions of softcommutation, the following conditions must be fulfilled: –
the switches must have threesegment characteristics;
–
the characteristics must be entirely described at each period;
–
the converter must include reversible source(s) able to provoke the conditions for spontaneous commutation of the switches at the right instant. The attractive properties of softswitching [9, 11] are:
–
the large reduction of switching losses;
–
the improved reliability due to reduced stress;
–
a limited frequency spectrum, which means an advantage with respect to EMI and losses in passive components;
–
a reduction of weight and volume of the components resulting from the higher switching frequency;
–
a higher bandwidth resulting from the high internal switching frequency;
–
integration of parasitic elements in the commutation mechanism (e.g. leakage inductance of the transformer in the resonant circuit).
The softcommutation domain has been a key research domain in power electronics in the last two decades. Numerous papers and conferences can be consulted for more information. It should be noted that all LHC power converters were designed with softcommutation topologies [12, 13, 14]. 11 Conclusions This paper has made an introduction and classification of the basic power converter components: sources and switches. From interconnection rules of the sources, the direct and indirect power converter topologies were deduced. A general and systematic method to synthesize was described and illustrated with examples. With the fast development of turnoff controllable power semiconductors, the commutation mechanism becomes more and more important. To improve the performance of converters, frequencies are increased with the minimization of losses and EMI perturbations. Local treatment of commutation is no longer possible and it is crucial to design a suitable topology for the commutation of highfrequency and highpower semiconductors. To reach these goals, it is necessary to create, through the circuit topology, the turnON and turnOFF conditions for the switches. Soft commutation is certainly the most appropriate and optimal solution. Acknowledgements For the work reported here, different references in the domain of power electronics and especially from publications of the Laboratoire d’Electrotechnique et d’Electronique Industrielle, Toulouse, France (LEEI) members were used. The first author would like to express his deep gratitude and admiration to Professor Foch, who initiated the work around the systematic synthesis of power converter topologies. It was an opportunity and honour to work with him for more than 10 years.
References [1] E. Carrol, in Proceedings of the CERNCAS Accelerator School on Power Converters for Particle Accelerators, Warrington, UK, 12 – 18 May 2004, edited by D. Brandt, CERN2006010 (CERN, Geneva, 2006), pp. 4356, DOI: http://dx.doi.org/10.5170/CERN2006010.43. [2] A. Rufer, in Proceedings of the CERNCAS Accelerator School on Power Converters for Particle Accelerators, Warrington, UK, 12 – 18 May 2004, edited by D. Brandt, CERN2006010 (CERN, Geneva, 2006), pp. 57102, DOI: http://dx.doi.org/10.5170/CERN2006010.57. [3] P. Viarouge, Passive devices magnetic, these proceedings. [4] F. Bordry, in Proceedings of the CERNCAS Accelerator School on Power Converters for Particle Accelerators, Warrington, UK, 12 – 18 May 2004, edited by D. Brandt, CERN2006010 (CERN, Geneva, 2006), pp. 275298, DOI: http://dx.doi.org/10.5170/CERN2006010.275. [5] D. Aguglia, IEEE Trans. Plasma Sci. 42, (2014) 3070. [6] H. Foch and J. Roux, Convertisseurs statiques d’énergie électrique à semiconducteurs, Patent FR 78324428, US patent 093106, 1978. [7] H. Foch, P. Marty and J. Roux, Use of duality rules in the conception of transistorized converters, PCI Proceedings., Münich 1980, pp 4B3 111. [8] Y. Cheron, H. Foch and T Meynard, Generalization of the resonant switch concept. Structures and performance, in Proc. Second European Conf. Power Electronics and Application – EPE, pp. 239–244, 1987. [9] Y. Cheron, Soft Commutation, Chapman & Hall, London, 1992, 233 pages. [10] R. Arches, F. Bordry, Y. Cheron, B. Escaut, H. Foch, P. Marty and M. Metz, Principes fondamentaux: Commutation dans les convertisseurs statiques. Techniques de l’Ingénieur, Traité de Genie Electrique, 9 (1989) D 3153. [11] K.H. Liu and F.C Lee, Zerovoltage switching technique in DCDC converters, IEEE Transactions on Power Electronics, Vol. 5, No. 3, July 1990, pp. 293 – 304. [12] F. Bordry, Power converters for particle accelerators: Development and production of the Large Hadron Collider (LHC) powering system, PCIM Conf., Nürnberg, 2005. [13] F. Bordry, G. Kniegl, V. Montabonnet, R. Pauls, H. Thiesen and B. Wolfes, Softswitching (ZVZCS) highcurrent, low voltage modular power converter, 9th European Conf. Power Electronics EPE’2001, Graz, 2001. [14] F. Bordry, V. Montabonnet, D. Nisbet, P. Korhonen, R. Turunen and H. Volotinen, Development, test and large production of softswitching high current power converters for particle accelerators, 11th European Conf. Power Electronics EPE’2005, Dresden, 2005. Bibliography N. Mohan, W. Robbins and T. Undeland, Power Electronics: Converters, Applications and Design (Wiley, Hoboken, 1989). M. Rashid, Power Electronics: Circuits, Devices and Applications (PrenticeHall International, Upper Saddle River, 1993). B. Bird, K. King and D. Pedder, An Introduction to Power Electronics (Wiley, Chichester, 1993). R.W. Eriksson and D. Maksimovic, Fundamentals of Power Electronics, 2nd ed. (Kluwer Academic Publishers, New York, 2001). H. Foch, P. Marty and Y. Cheron, in Proceedings of the CERNCAS Accelerator School on Power Converters for Particle Accelerators, Montreux, Switzerland, 26 – 30 March 1990, edited by T. Stuart, CERN1990007 (CERN, Geneva, 1990), pp. 3854, DOI: http://dx.doi.org/10.5170/CERN2006010.54.